The audience monitoring procedures currently used to calculate the size of a television and/or radio audience, which determines the price broadcasters can charge for advertising time, have come under growing criticism in recent years. The audience monitoring procedures typically depend on responses made by individuals who have agreed to be part of a sample audience. To calculate television ratings, for example, viewers are asked to log their time by pressing buttons on a "people meter" attached to their television sets. This method has come under attack by network executives who argue that many of these people simply fail to use the meters. Radio ratings are derived primarily from listeners' diaries, or written logs, and telephone surveys. Television stations could, of course, employ written logs or telephone surveys. However, written logs require even more discipline from listeners than electronic logs, and telephone surveys are vulnerable to faulty memories.
U.S. Pat. No. 4,955,070 (Welsh) describes a device that automatically identifies the station tuned in by a listener and logs the amount of time the listener remains within earshot. The device is based on "acoustic matching," and includes as its key components a microphone, a radio tuner and a micro-processor. The microphone detects sound near the person being monitored, and the microprocessor converts this sound into a digital code. The device then compares this code with a code similarly-derived from radio stations electronically monitored by the tuner. When the codes from the microphone and the tuner "match," the device logs the station and the amount of time the match continues. An important feature of this device is that, to determine whether there is a match, the respective signals being compared are first autocorrelated to determine two sets of correlation coefficients and then the two sets of correlation coefficients are compared. A disadvantage of the autocorrelation technique is that the correlation coefficients will be modified if the user distorts the frequency spectrum of the broadcast signal, e.g., by amplifying or suppressing selected frequency components.
French Patent No. 2,555,383 (Lucien) describes a similar audience monitoring system, i.e., one that digitalizes the respective signals being compared and then compares the resulting digital signals by way of a statistical correlation technique.
U.S. Pat. No. 4,388,644 (Ishman et al.), entitled "Apparatus for Monitoring a Multichannel Receiver," discloses a channel-monitoring system that includes means for injecting a signal into the antenna input of the receiver being monitored and means for detecting the injected signal at the output of the receiver. The monitor is capable of detecting the channel to which the receiver is tuned by varying the frequency of the injected signal over a prescribed set of frequencies that correspond to the channels to which the receiver may be tuned.
Thus, both the Welsh and Barrault techniques involve sophisticated digital signal processing to determine whether the respective signals match. The Ishman system requires a modification of the input signal of the receiver being monitored, which is believed to be too complex and expensive to be practical. A primary goal of the present invention is to provide an audience monitoring system that is simpler and more reliable than the known systems.